Using AI Where It Makes a Difference

Everyone has seen the amazing things that artificial intelligence can do, but this doesn’t necessarily translate into massive gains in productivity.

The issue isn’t that AI doesn’t work, the problem is where organizations are deploying AI. Using AI to compose emails or summarize meetings does little to speed organizational output.  As Eli Goldratt pointed decades ago, organizational productivity is throttled by the the throughput of the most constrained resource. Optimizing speed and processes before this bottleneck will not improve productivity. In fact, it may even make things worse by increase the queue and backlog at the constrained resource. ( See Theory of Constraints of Eliyahu M. Goldratt).

AI is great at producing documents, but is that the biggest constraint on your organization? Probably not. 

Organizations are great at producing documents and files. There is no shortage of documents inside any organization. Whether the documents are any good or you can find the documents when you need them is another matter. But the reality, is your organization is more likely constrained by organizational communication and decision making processes than by the need to create more documentations.

While decision-making is frequently delayed waiting for more information, the likelihood that the needed information is just one AI produced document away is virtually nil.  Organizational decision-making is more of an art than a science. It is a combination of timing, data, insight, experience, strategic intuition and risk tolerance of the decision stakeholders. The more important the decision seems, the more likely it is to involve more stakeholders and become exponentially more complicated.

AI can supply supporting documentation and make recommendations based on its research, but it is unlikely to supplant the human element in making strategic decisions. And as long as decision-making is on the critical path of organizational decision-making, AI is unlikely to make dramatic improvements in organizational productivity.

While reducing the cognitive burden on individuals for the mundane tasks like composing emails and summarizing meetings may free up some time to work on more important items, it is likely that the volume of automatically composed emails and meeting summaries will rush in to fill the cognitive burden vacuum.

Organizations that deploy AI at organizational bottlenecks will have far greater productivity gains than organizations that deploy AI tools anywhere that it is easy to deploy. Until AI is actually used  to address organizational bottlenecks like decision-making, Copilot buttons are just expensive status symbols that do little to boost overall productivity.